CN114498690A

CN114498690A - Multi-element composite energy storage optimal configuration method supporting large-scale renewable energy consumption

Info

Publication number: CN114498690A
Application number: CN202111657561.5A
Authority: CN
Inventors: 李学峰; 顾永正; 李国庆; 李秀芬; 李晓飞; 孔祥玉; 范晨亮; 刘磊; 张晋宇; 张振国
Original assignee: State Power Electric Power Inner Mongolia New Energy Development Co ltd
Current assignee: State Power Electric Power Inner Mongolia New Energy Development Co ltd
Priority date: 2021-12-30
Filing date: 2021-12-30
Publication date: 2022-05-13

Abstract

The invention relates to a multivariate composite energy storage optimal configuration method supporting large-scale renewable energy consumption, which researches a multivariate coordination ubiquitous scheduling control mode and a control strategy supporting source network charge storage aiming at energy storage system application scenes such as smooth output, planned power generation, peak clipping and valley filling, mixed scenes and the like; and (4) considering different types of application scenes and operation control strategies of the energy storage system, and researching the functional requirements and architecture design, data interaction, acquisition and functional support technology of the energy storage system. The method for stabilizing the wind power/photovoltaic active power fluctuation by using the energy storage system is researched. According to the characteristics of different types of energy storage equipment, a wind power/photovoltaic active power smooth control method based on wavelet packet decomposition and Empirical Mode Decomposition (EMD) is provided, the relation between a wavelet packet decomposition level and an empirical mode filtering order and energy storage energy is respectively researched, and a theoretical basis is provided for realizing multi-type energy storage energy division.

Description

Multi-element composite energy storage optimal configuration method supporting large-scale renewable energy consumption

Technical Field

The invention belongs to the technical field of energy storage system optimization, and particularly relates to a multi-element composite energy storage optimization configuration method supporting large-scale renewable energy consumption.

Background

Due to the influence of factors such as climate and geographical environment, wind energy has the characteristics of intermittent and random output, and the direct incorporation of the output power of the wind power plant into a power grid has great influence on the stability of a power system, the frequency of the power grid, the quality of electric energy, a power generation plan, dispatching and the like, so that the utilization of wind energy and the large-scale development of wind power are severely restricted. Therefore, the method has important practical significance on effectively stabilizing the problem of fluctuation of the output power of the wind power plant. Aiming at the problem, domestic and foreign scholars make active research and mainly focus on the following two ways, namely adjusting the output power of a fan by adjusting the pitch angle and changing the rotating speed of a generator to realize the purpose of smoothing the output power of a single wind generating set, however, the output power of the wind power plant units is possibly complemented and also possibly mutually superposed, so that the total output power of the wind power plant has larger fluctuation, and the effective utilization rate of wind energy is reduced to a certain extent; the other method is to consider the whole output power of the wind power plant, an energy storage system is arranged at the position of a grid-connected bus at the outlet of the wind power plant, and the handling capacity of the energy storage system is utilized to play a role in stabilizing the fluctuation of the wind power, namely when the wind power output suddenly rises, the energy storage device absorbs the power, otherwise, the power is output. Because the energy storage systems are various and have different characteristics, the energy storage systems are applied to different degrees in the aspect of power smoothing.

The reasonable configuration of the capacity of the energy storage system is an important guarantee for the economical and reliable operation of the new energy power generation system. There are two energy storage system capacity configurations currently in centralized and distributed form. Researches show that the capacity configuration of the centralized energy storage system can reduce the energy storage power and capacity and improve the economy of the system. The capacity configuration method of the energy storage system commonly used at present mainly comprises the following two methods: firstly, establishing an objective function by taking the minimum capacity of energy storage equipment as a target, and determining the energy storage capacity meeting the requirement, wherein the method has the limitation that the energy storage capacity configuration is only given from the technical angle, and the influence of the current higher energy storage cost on the economy of a photovoltaic/wind power system is not considered; secondly, establishing a target function with the minimum cost as a target, considering constraints such as power output and state of charge of the hybrid energy storage device, and providing the capacity of the energy storage system meeting the requirements. The disadvantage of this method is that the cost of energy storage is not calculated, and the cost of replacing and maintaining the equipment in the whole life cycle is not taken into account.

Disclosure of Invention

The invention aims to overcome the defects of the prior art, provides a multivariate composite energy storage optimal configuration method supporting large-scale renewable energy consumption, and researches a multivariate coordination ubiquitous scheduling control mode and a control strategy supporting source network charge storage aiming at energy storage system application scenes such as smooth output, planned power generation, peak clipping and valley filling, mixed scenes and the like; and (4) considering different types of application scenes and operation control strategies of the energy storage system, and researching the functional requirements and architecture design, data interaction, acquisition and functional support technology of the energy storage system.

The technical problem to be solved by the invention is realized by adopting the following technical scheme:

the multi-element composite energy storage optimal configuration method supporting large-scale renewable energy consumption comprises the following steps:

step 1, an energy storage planning model is constructed by considering an application scene and an operation control strategy;

step 2, performing optimal configuration on the stored energy fully consumed by the high-proportion renewable energy according to the model constructed in the step 1;

and 3, constructing an energy storage system function demand model according to the optimized configuration in the step 2, and solving through the model to obtain the multi-element composite energy storage optimized configuration method.

Further, the step 1 includes the steps of:

step 1.1, establishing a multi-objective optimization control strategy of the energy storage system;

and 1.2, constructing an energy storage planning model by considering the control strategy of the step 1.1.

Moreover, said step 1.1 comprises the steps of:

step 1.1.1, distributing energy of a hybrid system based on low-pass filtering;

the method of adopting the first-order low-pass filter converts the filter from the frequency domain to the time domain, and obtains the power of the battery and the super capacitor through the frequency division function:

wherein, P_bat(t) is the power of the battery at time t, λ is the filter coefficient, and its value range is 0 to 1, P_bat(t-1) is the power of the battery at the time t-1,

is a target total output value, P, of the hybrid energy storage system_cap(t) is the power of the super capacitor at the moment t, tau is a filtering time constant, delta t is a sampling interval, and meanwhile, the obtained filtering coefficient is in direct proportion to the battery power and in inverse proportion to the output of the super capacitor;

step 1.1.2, establishing a multi-objective optimization model control strategy according to the relation between the filter coefficient, the battery power and the output of the super capacitor;

the constraint conditions are as follows:

b₁≤λ(t)≤b₂

wherein, P_bat,eIs the rated power of the battery, C_soc,capIs SOC, C of a super capacitor_soc,medAt moderate level of SOC, b₁、b₂For factory-set parameters, M_capAnd for the energy storage capacity of the super capacitor, delta t is a charging and discharging time interval, and the constructed model is calculated through an NSGA-/(algorithm) to obtain a control strategy: dividing a command period of the output of the hybrid energy storage system into two parts, wherein in the ith command period T_iThe latter part of the period T_iWithin 2, at T _i2, ensuring the SOC of the super capacitor to be at a certain level, and selecting f₂The better solution is taken as the current solution; when the (i + 1) th instruction cycle is entered, the previous part of the cycle is T_i+1Within +1, the solution selected to obtain the better solution is taken as the current solution.

Moreover, the energy storage planning model of step 1.2 is:

the charging process is as follows:

C_SOC,bat(t)＝C_SOC,bat(t-Δt)+P_bat(t)Δtη_bat,c/C_bat

C_SOC,cap(t)＝C_SOC,cap(t-Δt)+P_cap(t)Δtη_cap,c/C_cap

the discharging process is as follows:

C_SOC,bat(t)＝C_SOC,bat(t-Δt)+P_bat(t)Δt/(η_bat,d·C_bat)

C_SOC,cap(t)＝C_SOC,cap(t-Δt)+P_cap(t)Δt/(η_cap,d·C_cap)

the constraint conditions are as follows:

|P_bat(t)|≤|P_bat,lim(t)|

|P_cap(t)|≤|P_cap,lim(t)|

wherein, C_SOC,bat(t) and C_SOC,cap(t) is the change condition of the SOC of the two energy storage systems; Δ t is the sampling interval; eta_bat,cThe charging efficiency of the battery; eta_cap,cIs the discharge efficiency of the cell; eta_bat,dThe charging efficiency of the super capacitor is improved; eta_cap,dIs the discharge efficiency of the supercapacitor; c_batIs the capacity of the battery; c_capIs the capacity of the supercapacitor; i P_bat,lim(t) | is the maximum charging power value allowed by the battery at the time t; i P_cap,lim(t) | is the maximum charging power value allowed by the super capacitor at the time t;

|P_bat,lim(t) | and | P_cap,lim(t) | is determined by the self characteristics and the residual energy of two energy storage systems, and the calculation method comprises the following steps:

|P_bat,lim(t)|＝min{P_bat,cmax,C_bat[C_SOC,bat,max-C_SOC,bat(t-Δt)]/(Δt·η_bat,c)}

|P_cap,lim(t)|＝min{P_cap,cmax,C_cap[C_SOC,cap,max-C_SOC,cap(t-Δt)]/(Δt·η_cap,c)}

and (3) discharging:

|P_bat,lim(t)|＝min{P_bat,dmax,C_bat[C_SOC,bat(t-Δt)-C_SOC,bat,max]η_bat,d/Δt}

|P_cap,lim(t)|＝min{P_cap,dmax,C_cap[C_SOC,cap(t-Δt)-C_SOC,cap,max]η_cap,d/Δt}

wherein, P_bat,cmaxThe maximum charging power value determined by the self characteristics of the battery; p_cap,cmaxThe maximum discharge power value is determined by the self characteristics of the battery; p_bat,dmaxThe maximum charging power value is determined by the self characteristics of the super capacitor; p_cap,dmaxThe maximum discharge power value is determined by the self characteristics of the super capacitor; c_SOC,bat,maxAn upper limit of the SOC constraint of the battery; c_SOC,cap,maxA lower constraint limit for the SOC of the battery; c_SOC,bat,minAn SOC constraint upper limit of the super capacitor; c_SOC,cap,minThe SOC constraint lower limit of the super capacitor.

Further, the step 2 includes the steps of:

step 2.1, calculating the equivalent cycle life of the battery;

calculating the discharge depth of the battery by a rain flow counting method, and calculating the equivalent cycle life of the battery according to the corresponding relation between the discharge depth and the cycle life of the battery;

step 2.2, calculating the rated power range of the energy storage system;

sampling by an interval of time, according to P_HESS(t)＝P_wf(t)-P_w(t) calculating the actually measured data and the prediction result of the wind turbine generator to obtain the power P of the hybrid energy storage system_HESSEstablishing an absolute power value | P of the hybrid energy storage system by adopting a nonparametric density estimation method_HESSA | distribution statistical model, and configuring the rated power of the hybrid energy storage system by using a kernel density estimation method;

step 2.3, calculating the capacity range of the energy storage system;

selecting an absolute value of the maximum energy value in one day, and solving an accumulative distribution function according to the probability density of the absolute value to obtain the capacity of the hybrid energy storage system;

step 2.4, calculating cost;

and 2.5, performing optimized configuration on the stored energy according to the calculation data of the steps 2.1 to 2.4.

Moreover, the cost calculated in step 2.4 includes the life cycle cost, the time value of capital, the construction stage of the project and the constraint condition;

the method for calculating the life cycle cost comprises the following steps: the life cycle cost of the hybrid energy storage system is mainly concentrated on the investment cost of the battery and the super capacitor module, the operation and maintenance cost after the system is operated, and the disposal cost after scrapping; the economic evaluation of the energy storage cost is influenced by the selection factors of the load characteristics, the parameters of the energy storage system and the working mode, and the economic evaluation of the energy storage system is carried out according to the applicable typical occasions of the power system and the technical specifications and parameter requirements of the energy storage system defined according to various power applications during calculation;

the time value of the fund is calculated by the following method:

wherein P is the time value, F is the final value, A is the annual value, and r is the interest rate; n number of device use periods;

the construction stage of the project comprises: a planning construction stage, an operation maintenance stage and a scrap disposal stage;

the calculation method of the planning construction stage comprises the following steps:

C_d＝n_b,p·P_bat+n_b,c·C_bat+n_c,p·P_cap+n_c,c·C_cap

wherein, P_bat，C_bat，P_cap，C_capRated power and rated capacity of the battery and the super capacitor respectively; n is_b,p,n_b,c,n_c,p,n_b,cThe power unit price and the capacity unit price of the battery and the super capacitor are respectively;

the calculation method in the operation and maintenance stage comprises the following steps:

C_m＝N·(m_bat·C_bat+m_cap·C_cap)

wherein m is_batAnd m_capMaintenance unit prices of the battery and the super capacitor respectively; n is the number of the service life of the equipment; q. q.s_batAnd q is_capThe replacement times of the battery and the super capacitor are respectively;

the calculation method of the scrap disposal stage comprises the following steps:

C_s＝l_b,p·P_bat+l_b,c·C_bat+l_c,p·P_cap+l_c,c·C_cap

wherein l_b,p、l_b,c、l_c,pThe power disposal unit price and the capacity disposal unit price of the battery and the super capacitor are respectively;

the constraint conditions include: node power balance constraint, node power fluctuation constraint, node voltage constraint, branch power flow constraint, energy conservation constraint, confidence constraint and SOC constraint.

Furthermore, the energy storage optimization configuration in step 2.5 is:

according to the principle of wind power ratio exceeding the comprehensive target domain range, the wind power ratio is counted by adopting the frequency repetition method probability of the overflow target domain, according to the energy storage response characteristic, the large wind power ratio is compensated by a storage battery, the small wind power ratio is compensated by a super capacitor, and the respective running time, rated power and capacity ratio of the hybrid energy storage are determined.

Further, the step 3 includes the steps of:

step 3.1, calculating the relation between the energy storage capacity demand and the renewable energy source regulation;

the controller collects the output power P of the photovoltaic module_sSetting the value P of the power to be connected to the power through a first-order digital low-pass filtering algorithm_outAnd calculating the force output value of the energy storage system:

P_b＝P_out-P_s

the required energy storage system capacity is:

selecting photovoltaic power station output curve samples with a time interval, continuously selecting and accumulating sample capacity from 1 day one by one, calculating a capacity configuration result of the photovoltaic power station under the sample length, drawing a frequency distribution histogram of the number of days of the samples, and performing normal distribution fitting on the frequency distribution histogram to obtain an energy storage capacity demand and renewable energy source regulation relation;

step 3.2, calculating the output data characteristics of the photovoltaic power station;

the photovoltaic power station output data characteristics comprise autocorrelation of output data and similar daily clustering of the output data of the photovoltaic power station; wherein the autocorrelation of the contribution data is;

wherein cov (-) is covariance, var (-) is variance, p_xyIs a correlation coefficient, x_tData value at any moment, y, of the photovoltaic output data_tIntercepting a data sample for calculating by taking 100 days as a span when photovoltaic output data are obtained for historical data and analyzing;

the similar daily clustering of the photovoltaic power station output data is as follows: selecting the solar irradiation intensity, the irradiation time and the air temperature as clustering indexes of photovoltaic output to obtain daily characteristic vectors:

wherein the content of the first and second substances,

the maximum value, the average value in the morning and the average value in the afternoon of the solar irradiation intensity on the ith day;

the maximum air temperature, the average air temperature in the morning and the average air temperature in the afternoon of the ith dayAir temperature; t is the irradiation time length of the ith day; according to the autocorrelation of the output data, selecting photovoltaic data with the span of 100 days for clustering analysis to obtain strong correlation between the photovoltaic output and weather conditions;

3.3, calculating the span based on the optimal sample capacity estimation;

the optimal sample volume is:

where n is the sample volume, σ²Is the standard deviation of the sample, t_αThe method is characterized in that the method is a standard normal distribution bilateral quantile, u is an overall mean value, epsilon is relative precision, and a span based on the optimal sample capacity estimation is obtained: with the reduction of the allowable error precision value, the optimal sample capacity estimation of each weather category is increased, and the overall data span is continuously increased.

The invention has the advantages and positive effects that:

aiming at energy storage system application scenes such as smooth output, planned power generation, peak clipping and valley filling, mixed scenes and the like, a comprehensive scheduling control mode and a control strategy supporting source network load storage multi-element coordination are researched; and (4) considering different types of application scenes and operation control strategies of the energy storage system, and researching the functional requirements and architecture design, data interaction, acquisition and functional support technology of the energy storage system. The invention utilizes an energy storage system to stabilize the fluctuation of the wind power/photovoltaic active power to carry out research. According to the characteristics of different types of energy storage equipment, a wind power/photovoltaic active power smooth control method based on wavelet packet decomposition and Empirical Mode Decomposition (EMD) is provided, the relation between a wavelet packet decomposition level and an empirical mode filtering order and energy storage energy is respectively researched, and a theoretical basis is provided for realizing multi-type energy storage energy division. Meanwhile, the invention provides an energy storage capacity configuration method based on a wind power prediction error model, and reasonable matching of energy type and power type energy storage systems is realized.

Drawings

FIG. 1 is a schematic diagram of a Pareto frontier and an optimal solution of the present invention;

FIG. 2 is a schematic diagram of the hybrid energy storage system compensating for deviations between the measured value and the predicted value of the wind power;

FIG. 3 is a schematic of the charge and discharge cycle of the present invention, (a) the charge half cycle; (b) a discharge half cycle period; (c) a cycle of charging completion; (d) a complete cycle period of discharge;

FIG. 4 is a schematic diagram of a rain flow counting method according to the present invention, (a) SOC versus time curve; (b) rotation of the time-varying curve of SOC by 90 ° (c) cycle period; (d) a cycle period; (e) a cycle period; (f) three cycle periods and one half cycle period;

FIG. 5 is a schematic diagram of a cycle of calculating SOC of a battery according to the rain flow counting method of the present invention;

FIG. 6 is a graph of cycle life versus depth of discharge for the present invention;

FIG. 7 shows | P of the present invention_HESSA histogram of |;

FIG. 8 shows | P of the present invention_HESSA probability density function graph of l;

FIG. 9 shows | P of the present invention_HESSA | probability density function graph;

FIG. 10 is a schematic daily maximum absolute energy diagram of the hybrid energy storage system of the present invention;

FIG. 11 shows | E of the present invention_maxA cumulative distribution function of l;

FIG. 12 is a schematic life cycle segment division of the hybrid energy storage system of the present invention;

FIG. 13 is a graph of the predicted annual average cost of HESS energy storage as a function of battery rated capacity;

FIG. 14 is a graph of BESS energy storage average-year-cost as a function of battery rated capacity in accordance with the present invention;

FIG. 15 is a schematic diagram of the photovoltaic power plant filter control algorithm of the present invention;

FIG. 16 is a data span histogram corresponding to the energy storage maximum capacity requirement value of the present invention;

FIG. 17 is a graph of a typical output for five days according to the present invention;

FIG. 18 is a weather-based photovoltaic data clustering plot of the present invention;

FIG. 19 is a diagram illustrating the estimation of the most suitable sample volume according to the present invention.

Detailed Description

The present invention will be described in further detail with reference to the accompanying drawings.

The multi-element composite energy storage optimal configuration method for supporting large-scale renewable energy consumption is characterized by comprising the following steps of: the method comprises the following steps:

step 1, an energy storage planning model is constructed by considering application scenes and operation control strategies.

Step 1.1, establishing a multi-objective optimization control strategy of the energy storage system.

Step 1.1.1, hybrid system energy is distributed based on low-pass filtering.

In order to fully exert the respective advantages of energy type and power type energy storage, from the perspective of reducing investment and operation cost, the internal energy distribution rule of the hybrid energy storage is that a super capacitor with high power density and long cycle life follows the principle of preferential charging and discharging, and simultaneously serves as a power buffer to execute high-frequency components in a power instruction of the hybrid energy storage, so that the battery is prevented from being charged and discharged in a small cycle and super-multiple mode, the service life of the battery is prolonged, and the operation cost is reduced; the battery with high energy density is a main energy source in the hybrid energy storage system, is responsible for bearing a low-frequency component part for a long time, and can adjust the charge state of the capacitor in real time and respond to a high-frequency part of a next command in time.

The method adopts a first-order low-pass filter method to divide the deviation Per between the predicted value and the measured value of the wind power into a high part and a low part. The filter transfer function g(s) is:

wherein s is a proprietary symbol in the transfer function; tau is a time constant of the filtering,

in order to facilitate the optimization calculation, the filter is converted from a frequency domain to a time domain, and the power of the battery and the super capacitor is obtained through the frequency division effect of the filter:

is a target total output value, P, of the hybrid energy storage system_capAnd (t) is the power of the super capacitor at the moment t, tau is a filtering time constant, and delta t is a sampling interval.

The value range of the filter coefficient is 0-1, the smaller the filter coefficient is, the flatter the power output of the battery is, and meanwhile, the power of the super capacitor fluctuates more severely up and down; if the filter coefficient takes a larger value, the flatness degree of the output power of the battery is reduced, the output power is integrally higher, the task borne by the corresponding super capacitor is reduced, and the fluctuation output is also reduced. The conclusion is drawn that the energy distribution of the hybrid energy storage can be determined by controlling the filter coefficient lambda, and the filter coefficient is in direct proportion to the battery power and in inverse proportion to the output of the super capacitor;

step 1.1.2, establishing a multi-objective optimization model control strategy according to the relation between the filter coefficient, the battery power and the output of the super capacitor.

In the tracking control of the hybrid energy storage system, how to reasonably determine the value of lambda so as to fully exert the maximum efficacy of the two types of energy storage can be solved through an optimization theory, and the premise of using the optimization theory is to establish an optimization model.

In the tracking control, the service life of the stored energy is analyzed from the perspective of the service life of the stored energy, the service life of the stored energy is restricted by factors such as cycle times, depth of discharge (DOD) and the like, particularly, the cycle life of the stored energy of the battery is low, the overcharging and deep discharging of the stored energy of the battery are avoided, the output power amplitude of the battery at each sampling point is limited, and the output power amplitude is taken as one of targets. On the other hand, due to the intervention of the super capacitor, the characteristics of high energy density and high cycle number can be used for bearing a high-frequency component part with a high amplitude in a short time, in order to meet the requirement that the super capacitor can have a strong high-frequency power output capacity at any time, the SOC of the super capacitor is kept at a relatively moderate level, the requirement of charging or discharging at the next moment can be met, and the requirement is taken as a second optimization target.

When the number of the optimization targets is two or more, the optimization problem belongs to a multi-target optimization problem. The multi-objective optimization problem can be described as follows:

wherein f (x) is an objective function to be optimized, wherein f (x) is f₁(X),f₂(X),…,f_p(X)]^T(ii) a X is the variable to be optimized, wherein X ═ X₁,x₂,…,x_n)^T；g_i(X) is a constraint condition; v-min is vector optimization, i.e., each sub-target in the vector objective function F (X) is equally minimized.

According to the requirements of the optimization target, the established model for tracking optimization control of the hybrid energy storage system is as follows:

optimizing an objective

Wherein, C_soc,medThe SOC is a moderate level, and generally takes about 0.5; p_bat,cIs the rated power of the battery; delta t is a charging and discharging time interval; m_capThe energy storage capacity of the super capacitor; f. of₁Minimizing the depth of discharge at that time for the limitation of the battery power amplitude; f. of₂To minimize super-capacitorCurrent SOC value and C_soc,medI.e. to ensure that the supercapacitor SOC is at a moderate level to cope with the higher frequency output at a future moment.

In the energy distribution of mixed energy storage, the pair P can be determined due to the fact that the filter coefficient lambda plays a role in determining_bat,tAnd P_cap,tThe problem of optimization is transformed into a problem of optimization for lambda, and the optimization objective is transformed into:

the constraint conditions are as follows:

b₁≤λ(t)≤b₂

wherein, P_bat,eIs the rated power of the battery, C_soc,capIs SOC, C of a super capacitor_soc,medAt moderate level of SOC, b₁、b₂For factory-set parameters, P_bat(t-1)，

And C_soc,cap(t-1) can be obtained by calculation from the correlation formula. P_bat,c，M_cap，C_soc,mod，Δt，b₁And b₂Is a set parameter.

The multi-objective optimization problem does not have the absolute optimal solution which enables all the targets to reach the optimal simultaneously, wherein all the targets are always in a conflict state, and the improvement of one objective function needs to be achieved at the cost of the reduction of the other objective function. For the minimum value multi-objective optimization problem minf (), the Pareto optimal solution is defined to be in the feasible domain of the design variable, and for the variable X, if and only if no other variable X exists^*Satisfy f without violating the constraint_i(X)<f_i(X^*) At least one i is present such that f_i(X)<f_i(X^*) If yes, the variable X is called as a non-dominant solution, namely a Pareto optimal solution. The Pareto optimal solutions are not unique, a plurality of Pareto optimal solutions form a Pareto optimal solution set (also called a Pareto frontier or non-dominated solution set), and solutions formed by the Pareto optimal solutionsCalled Pareto frontier, as shown in fig. 1.

FIG. 1 shows a solution space for an unconstrained optimization problem with two objectives. In the figure, two targets f due to point E₁、f₂Are smaller than the R point, so the R point is not the optimal solution. Similarly, all solutions that are within the feasible region and not on the boundary ABECD are not optimal solutions. All solutions on the curve ABECD are Pareto solutions, the whole curve forms a Pareto solution set, namely a Pareto leading edge, and the task of multi-objective optimization is to find the curve.

The NSGA/algorithm not only has all the advantages of the evolutionary algorithm, but also has the characteristics of no restriction of search function continuity, high efficiency of parallel computation and better universality, so the algorithm is selected to solve the multi-objective optimization problem.

The non-dominated classification process of the NSGA algorithm embodies the high efficiency of the NSGA algorithm, and simplifies the multi-objective optimization problem into a form of fitness function. The method can solve the optimization problem of any target number and solve the problem of maximum or minimum value. The NSGA/algorithm is a fast non-inferior ranking method improved on the basis of the NSGA method, which defines the crowding distance and estimates the density of solutions around a certain point to replace the adaptive value sharing. The NSGA/algorithm can effectively overcome the defects of NSGA, and the calculation complexity is from 0 (mM)³⁾Down to 0 (mM)²) And has an optimal reservation mechanism and does not need to determine a sharing parameter. The non-inferior solutions obtained by the algorithm are uniformly distributed on a target space, have good convergence and robustness and become evolution multiple targets.

The multi-objective optimization model control strategy is as follows: dividing a command period of the output of the hybrid energy storage system into two parts, wherein in the ith command period T_iThe latter part of the period T _i2, in order to prevent the situation that the predicted value of the wind power in the next command period may be changed greatly, and the target output power of the hybrid energy storage is also changed greatly, the T is required to be set_i2, the SOC of the super capacitor is ensured to be at a certain level to cope with the high-frequency component of the hybrid energy storage power, so f is selected₂The better solution isThe current solution; when the (i + 1) th instruction cycle is entered, the previous part of the cycle is T_iWithin +11, the output power amplitude of the battery is reduced as much as possible under the condition that the super capacitor is responsible for bearing a part of high-frequency components so as to prolong the service life of stored energy, and therefore a better solution is selected as the current solution.

The intervention of a large-scale energy storage system can improve the capacity of adjusting the output power of wind power in real time to track a power generation plan. How to utilize the energy storage system and realize the optimal scheduling of power between the two is one of the key problems of solving the wide application of large-scale energy storage systems in new energy power generation. The basic control idea is that according to the actual output power of the wind power plant and the power generation instruction target issued by the power grid, the energy storage system achieves the purpose of reducing or eliminating the deviation of the actual output power and the power generation instruction target through charging or discharging. When the power generation instruction value is larger than the actual output power, the energy storage system needs to release energy to compensate the shortage of the wind power; and when the power generation instruction value is smaller than the actual output power, the energy storage system needs to absorb excessive power and store the energy.

Step 1.2.1, power type and energy type characteristic analysis.

Taking a lithium ion battery and a super capacitor as examples, the super capacitor belongs to a power type energy storage device, has large output power variation range, high variation rate and multiple charging and discharging cycle times, and is mainly used for bearing high-frequency components of power; the lithium ion battery belongs to an energy type energy storage device, has small output power variation range, slow variation rate and few charging and discharging cycle times, and is mainly used for bearing low-frequency components of power.

The power density of the super capacitor can reach 2 times or even more than 10 times of that of the lithium ion battery, the cycle times of the super capacitor are far larger than that of the lithium ion battery, the characteristics of high power density, long cycle life and short charging and discharging time of the super capacitor are highlighted, and the energy density of the lithium ion battery is far higher than that of the super capacitor, so that the super capacitor has the capability of continuously and stably charging and discharging for a long time.

In order to compensate the deviation between the measured value and the predicted value of the wind power, the power of the hybrid energy storage system is as follows:

P_HESS(t)＝P_wf(t)-P_w(t)

wherein, P_HESS(t) is the power of the hybrid energy storage system at time t, P_wf(t) is the predicted value of wind power, P_w(t) is the measured value of wind power, and if P is found by the formula_w(t)＞P_wf(t), charging the hybrid energy storage system when the measured value of the wind power is larger than the predicted value; if P_w(t)＜P_wf(t), when the measured value of the wind power is smaller than the predicted value, discharging the hybrid energy storage system; if P_w(t)＝P_wfAnd (t), stopping the operation of the hybrid energy storage system when the measured value of the wind power is equal to the predicted value. The hybrid energy storage system makes up the schematic of the deviation between the wind power measured value and the predicted value, as shown in fig. 2.

In fig. 2, the solid line represents the total power of the hybrid energy storage system, and two points with a time interval Δ t (which is a shorter time) are taken, and the power of the hybrid energy storage system is P_HESS(t-. DELTA.t) and P_HESS(t) of (d). As indicated by the notation in FIG. 2, the battery power at time t- Δ t is P_bat(t- Δ t) power of the supercapacitor is P_cap(t- Δ t), the power of the battery and the super capacitor at time t being P respectively_bat(t) and P_cap(t) of (d). After a time of Δ t, there are

Because the super capacitor has high power density, higher power can be output in a shorter time. Thus, FIG. 2 can derive Δ P_cap＞ΔP_batThe short-time change rate of the super capacitor is larger than that of the battery, the super capacitor is suitable for the situation that the measured value and the predicted value of the wind power are increased suddenly instantly, and the battery is suitable for compensating the error that the measured value and the predicted value of the wind power are stable due to larger energy density.

The energy-type and power-type energy storage systems described in fig. 2 differ in the ability to release or absorb power at different time scales. Therefore, the concept of the rate of change of the energy storage system output is introduced to distinguish the capabilities of the two types of energy storage media.

The limitation on the output change rate of the energy storage system means that the output change amount of the power change rate of the energy storage system in a period of time does not exceed a certain limit value in the operation process of the energy storage system, and the definition is as follows:

wherein v is_bat,maxAnd v_cap,maxThe maximum power change rate of the battery and the supercapacitor, respectively. From the above analyzed characteristics, v is known_bat,max≤v_cap,max。

Under the influence of respective capacity bearing ranges, the energy type energy storage system is difficult to control the high-frequency component of the wind power prediction error due to the factors of low power density, short charging and discharging time, limited cycle life and the like; the power type energy storage system is difficult to undertake the low-frequency regulation and control of the wind power prediction error for a long time due to low energy density. Therefore, the hybrid energy storage system formed by the battery and the super capacitor has the advantages of high energy density, high power density, long cycle times and the like, and the problem that the power type or energy type energy storage system which is used independently is restricted by the factors of the energy density, the power density, the service life and the like is solved to the greatest extent.

And 1.2.2, establishing an analysis model of the hybrid energy storage system.

The mathematical model of the hybrid energy storage system is a mathematical expression which satisfies given constraints and can describe the charging and discharging processes of the hybrid energy storage system. The power value of the hybrid energy storage system is the deviation of the predicted value and the measured value of the wind power, the deviation is positive and represents discharge, the deviation is negative and represents charge, and the deviation is zero and represents that the system stops running. And respectively marking the change conditions of the SOC of the two energy storage systems as C after the output of the energy storage system at the moment t_SOC,bat(t) and C_SOC,cap(t) has the following relationship:

the charging process is as follows:

C_SOC,bat(t)＝C_SOC,bat(t-Δt)+P_bat(t)Δtη_bat,c/C_bat

C_SOC,cap(t)＝C_SOC,cap(t-Δt)+P_cap(t)Δtη_cap,c/C_cap

the discharging process is as follows:

C_SOC,bat(t)＝C_SOC,bat(t-Δt)+P_bat(t)Δt/(η_bat,d·C_bat)

C_SOC,cap(t)＝C_SOC,cap(t-Δt)+P_cap(t)Δt/(η_cap,d·C_cap)

the constraint conditions are as follows:

|P_bat(t)|≤|P_bat,lim(t)|

|P_cap(t)|≤|P_cap,lim(t)|

and (3) discharging:

And 2, carrying out optimal configuration on the stored energy fully consumed by the high-proportion renewable energy according to the model constructed in the step 1.

The energy storage capacity configuration method based on the wind power prediction error model is provided through analyzing the power characteristics of renewable energy sources in a certain place and energy storage application scenes under different conditions, and reasonable matching of energy type and power type energy storage systems is achieved.

On the basis of the hybrid energy storage power distribution method, a wind power-hybrid energy storage system formed by a wind power-storage battery/a super capacitor is taken as a research object, the complementary characteristics of the storage battery and the super capacitor are utilized, the full-period service life quantitative calculation is fully considered, and a capacity configuration model of the hybrid energy storage system is established. According to the method, the annual low cost in the operation period is taken as a target function, and constraint conditions such as fluctuation, confidence coefficient, energy storage output and SOC of wind power grid-connected power are considered.

And 2.1, calculating the equivalent cycle life of the battery.

The cycle life of the super capacitor is influenced by factors such as temperature, discharge rate and working current waveform, and the influence degree does not have concise quantitative indexes, and meanwhile, the service life of the super capacitor is as long as more than ten years, the cycle life is as long as 50-100 ten thousand times and is far longer than the service life of a battery, so the replacement time interval of the super capacitor is set to be constant. The following description focuses on the calculation method of the equivalent cycle life of the battery.

The life of the battery is closely related to the operating mode, and the larger the depth of discharge (DoD), the shorter the cycle life. The method comprises the steps of firstly calculating the discharge depth of a battery by a rain flow counting method, and then calculating the equivalent cycle life of the battery according to the corresponding relation between the discharge depth and the cycle life of the battery.

Fig. 3 shows a schematic of the charge and discharge cycle.

Wherein a complete cycle period is composed of a discharge half-cycle period and a charge half-cycle period, i.e. the cycle period is C_SOC,a→C_SOC,b→C_SOC,aWith a charge/discharge half-cycle period of C_SOC,a→C_SOC,bDepth of discharge D_DoD＝|C_SOC,a-C_SOC,bL, where 0 is equal to or less than C_SOC,a≠C_SOC,b≤1。

In engineering practice, the output power of the energy storage for stabilizing the wind power fluctuation is usually non-constant power, the change of the battery SOC is generally irregular, and a clear charging and discharging cycle sequence cannot be directly marked. Therefore, the problem to be solved is how to divide the SOC variation curve to obtain a charging and discharging cycle sequence with clear physical significance.

Rain flow counting method:

the rain flow counting method can be called as a tower top method, is mainly applied to the engineering field, and is particularly widely applied to the calculation of the fatigue strength and the service life of mechanical devices. The main function of the method is to study the non-linear counting relation between strain and time, namely to determine a set of non-periodic cycles of the data samples by using a rain flow counting method. The principle of calculating the battery discharge depth based on the rain flow counting method is as follows:

(1) the SOC-time curve is rotated 90 ° clockwise, marking the start point of the data and each peak point relative to the inner edge.

(2) Raindrops fall from the eave like, a vertical line is vertically drawn along the peak point of the curve until the rain flow encounters a new peak value larger than the original peak value or a new valley value smaller than the original valley value.

(3) When the rain stream encounters rain falling from a high roof, the flow will stop and constitute a cycle.

(4) Each cycle is plotted according to the rain stream starting and stopping points, and the cycle is read out completely while the peak-to-valley value is recorded.

(5) The difference in the lateral peak-to-valley distance for each sub-period is taken as the depth of discharge for that cycle.

The process of calculating the depth of discharge according to the SOC of the battery by the rain flow counting method is shown in FIG. 4, wherein a black solid line (0-1-2-3-5-6-7) represents a change curve of the SOC of the battery, a t axis represents time, and an x axis represents an SOC amplitude. The 300 time-varying curve of fig. 4(a) is rotated clockwise by 90 °, and the resulting curve is shown in fig. 4 (b). Making a straight line from each peak point downwards, and ending when the straight line intersects with the curve, two cycle periods can be obtained: 1-2-1' and 5-4, as shown in FIG. 4 (c). And (4) eliminating the two calculated cycles, and continuously searching whether a closed cycle still exists to obtain the SOC curve in the figure 4(d), namely, a cycle period exists: 3-6-3'. By eliminating this cycle, it can be seen from FIG. 4(e) that the SOC curves remain for only one half cycle period 0-7. In conclusion, three cycle periods (1-2-1', 5-4', and 3-6-3') and one half cycle period (0-7) are obtained as shown in fig. 4(f) through calculation by the rain flow counting method.

Therefore, the SOC of the battery obtains a series of cycle periods and half cycle periods with known discharge depths by a rain flow counting method, and a foundation is laid for the subsequent calculation of the equivalent cycle life of the battery.

Calculating the battery discharge depth by a rain flow counting method:

for the battery SOC of each wind power plant in the time period of 06:00-08:00 on a certain day, a cycle period process is calculated by adopting a rain flow counting method, and the process is shown in FIG. 5. For the original SOC curve of the battery shown in fig. 5(a), the extreme point is shown in fig. 5(b), and the position of the extreme point is the time when the charging and discharging state of the battery changes. Taking the maximum point 10 as an example, within the time period of the maximum point 9 to 10, C_SOCFrom 0.4789 up to 0.5062, the battery is charged; in the time interval of the extreme point 10 to 11, C_SOCFrom 0.5062 down to 0.4846, the battery discharges. Fig. 5(c) shows 8 cycle periods and 2 half cycle periods obtained by calculating the SOC of the battery, and the depth of discharge of each cycle is shown in table 1.

TABLE 1 cycle periods and depth of discharge

Calculating the equivalent cycle life:

after a period of use, the battery gradually enters a decline period, and the prominent performance is that the capacity is rapidly reduced. According to the requirement of the battery use maintenance standard, if the actual capacity of the battery is reduced to be below 80% of the rated capacity, the battery pack is considered to reach the service life, and a spare battery is arranged to be replaced and discarded. The cycle life of a battery is closely related to the depth of discharge, and usually, a battery manufacturer will calibrate the cycle life of the battery at different depths of discharge as a reference value for calculating the cycle life, for example, the correspondence between the depth of discharge and the cycle life of a certain type of battery is shown in table 2.

Because the battery corresponds to different cycle lives at different depths of discharge, the cycle lives at different depths of discharge need to be unified to a certain reference value (usually, the depth of discharge is selected to be 100%) for quantitative comparison, i.e., the equivalent cycle life is calculated. The cycle life of a battery is affected by many factors, such as temperature, peak current, and the number of charge and discharge cycles. To simplify the calculation, only the influence of the number of charge and discharge times and the depth of discharge on the cycle life of the battery is considered here.

TABLE 2 corresponding relationship between discharge depth and cycle life of certain type of battery

In practical application, the battery is not limited to the depth of discharge shown in table 2, and curve fitting can be performed according to the data in table 2 to obtain a functional relation curve corresponding to the depth of discharge and the cycle life. Common fitting methods include an N-order function method, a power function method, a piecewise fitting method and the like, a fourth-order function is adopted to represent the relation between the cycle life and the discharge depth, and a corresponding curve is shown in FIG. 6.

Defining the discharge depth of the ith cycle period of the storage battery as D_oDiThen the equivalent cycle life is

Wherein N is_tcf(D_oD1) Is the corresponding cycle life when the depth of discharge is 1; n is a radical of_tcf(D_oDi) When the depth of discharge is D_oDiCycle life of the time. The equivalent cycle life of the battery over the duty cycle is:

define the battery life loss as:

when N is equal to N_tcf(D_oD1) Or T is 1, it is considered that the battery pack of the batch has run out of life, and the battery pack needs to be replaced.

Taking the cycle period and the depth of discharge shown in table 2 as an example, the equivalent life of the battery is calculated, and the number of times of equivalent complete discharge of the battery in this operation period is L of 0.9805, and the life loss is T of 0.0018.

And 2.2, calculating the rated power range of the energy storage system.

As described in step 2.1, a wind power prediction error analysis is necessary when determining the rated power of the energy storage system. Establishing a distribution model of the prediction error is also the process of establishing an output power distribution model of the energy storage system. According to the given wind power combination result, the sampling time interval is 15 min. The measured data and the prediction result of the wind turbine generator are processed according to P_HESS(t)＝P_wf(t)-P_w(t) calculating to obtain the power P of the hybrid energy storage system_HESS. When the rated power is determined, the absolute value | P of the power of the hybrid energy storage system is obtained because the charging and discharging actions of the energy storage system are bidirectional_HESSThe analysis is performed.

To | P_HESSThe statistical analysis is performed, and the histogram obtained by statistics is shown in fig. 7. In FIG. 7, | P is reflected_HESSAnd the | height is concentrated within 10MW, and in order to comprehensively describe the distribution rule of the | height, a non-parameter density estimation method is adopted to establish a distribution statistical model of the | height.

And configuring the rated power of the hybrid energy storage system by using a nuclear density estimation method. Taking the kernel function as a gaussian kernel function, taking n as 100, calculating to obtain that the bandwidth h of the gaussian kernel function is 0.4189, and calculating results are shown in fig. 8 and 9 as | P respectively_HESSProbability density function of |(PDF) and Cumulative Distribution Function (CDF).

Solid line | P in FIG. 8_HESSThe probability density function of | can be intuitively seen that the probability density function can describe | P more smoothly_HESSOverall distribution characteristics of l. The cumulative distribution function of the density function is then calculated, and the result is shown in fig. 9.

The point marked in fig. 9 satisfies the requirement of compensating for the predicted power deviation with a probability of 98%, and the corresponding abscissa value of 23MW is the rated power of the hybrid energy storage system.

And 2.3, calculating the capacity range of the energy storage system.

In a period of time, the energy of the energy storage system can be obtained by integrating the power of the energy storage system in time, in order to determine the capacity of the energy storage system, the absolute value of the maximum value of the energy in one day, namely the maximum absolute energy in each day, is taken and is recorded as | E_maxL. FIG. 10 is | E for a time period of 250 days_maxThe | curve.

The kernel density function in the upper graph is a Gaussian kernel function, n takes a value of 100, and the bandwidth of the Gaussian kernel function is calculated to be 1.0595. It can be seen that the estimated probability density curve not only has better smoothness, but also does not ignore the main peak value, and meets the basic requirement for describing the distribution characteristics, and the probability density finds its cumulative distribution function, and the result is shown in fig. 11.

The points marked in fig. 11 are daily requirements for meeting 98% of the energy storage capacity, and the abscissa values corresponding to the points marked in the figure are the capacity of the hybrid energy storage system, and the rated capacity of the hybrid energy storage system is 30MWh, which is denoted as M-30 MWh.

And 2.4, calculating the cost.

The calculated cost includes the life cycle cost, the time value of capital, the construction stage of the project and the constraint condition;

(one) full life cycle cost

In the traditional engineering project management, only the cost of an engineering construction stage is generally considered, and the later additional cost generated after the project is operated is usually ignored. The method does not well reflect the economy of the project in the whole operation period, so a life cycle cost analysis (LCC) method is introduced to analyze the cost of the whole life cycle of the hybrid energy storage system.

The life cycle cost refers to the total expense of the equipment, system or project in the whole process of planning, designing, manufacturing, purchasing, installing, operating, maintaining, updating and scrapping, considering the total expense from the long-term economic benefit of the equipment, system or project.

The management of the whole life cycle is a complex project, and is different from other management concepts in that:

(1) the life cycle management is a system project: to achieve each of the staged goals and to achieve the optimum of the ultimate goal (including economic value of investment, social impact and environmental benefits) it is necessary to rely on a systematic and scientific management approach.

(2) The life cycle management is a continuous project: the management behavior is expressed in the whole project construction process, on one hand, different characteristics and targets exist in different stages, the management behavior is staged, and meanwhile, the management behavior requires that working rings in all stages are connected in a ring mode, so that the management behavior has good integrity.

(3) The life cycle management is a multi-party project: all participating main bodies have unified interest appeal and conflict, and all the participating main bodies are required to cooperate uniformly and supervise independently.

The life cycle cost of a hybrid energy storage system is mainly focused on the investment cost of the battery and the supercapacitor module, the operation and maintenance cost after the system is operated, the disposal cost after scrapping and the like. The economic evaluation of the energy storage cost is affected by factors such as load characteristics, selection of parameters and working modes of the energy storage system (independent operation, combined operation in a sleep mode, and the like), and therefore, the economic efficiency of the energy storage system should be evaluated according to typical situations of the applicable power system and according to technical specifications and parameter requirements of the energy storage system defined by various power applications.

Time value of capital

Due to the timeliness of the funds, the value of the same funds at different time nodes also differs, which results in the failure to directly compare the amount of funds occurring at different time points. Considering that the life cycle of the power equipment is as long as decades, the time value of capital must be considered when analyzing the cost of the power equipment, and cash flows generated at different time points should be converted to standard time points for comparison so as to have comparability in time.

In engineering economy, calculating the time value of a fund generally involves the following three concepts:

(4) present value p (present value), the value at which funds are generated (or converted) to the beginning of the entire calculation cycle, typically represents principal in a time value calculation. In engineering economy, the present value is usually used to represent the value of the investment at the initial moment.

(5) The final value, F (future value), is the value that funds have incurred at (or converted to) the end of the entire calculation cycle, and in time value calculations typically represents the sum of the instinct. In engineering economics, the end value is typically used to represent the value of the investment or revenue at the time the project terminates.

(6) The annual value a (annual value), the value of the equity funds recurring during each sub-period of the overall calculation cycle. In engineering economics, the annual value is typically used to represent the value of an equal amount of capital that is paid out or in every other time.

Since the above 3 time values occur at different time points and need to be compared by converting the time values into one form, the present value P is used as the main expression form of the time value. The conversion formula of the final value F, the annual value A and the current value P is

Wherein r is the interest rate; n number of device use periods, usually in units of years.

(III) construction stage of engineering

The method is applied to capacity configuration of the hybrid energy storage system, cost of each stage of the whole operation life cycle of the hybrid energy storage system is used as a research object, the lowest total cost is used as an optimized target function, and meanwhile, each technical index of energy storage system operation is used as a constraint to find an optimal configuration scheme which meets economic and grid-connected requirements at the same time.

The service operation period of the electric power engineering project is long, the whole life cycle cost management exists in each stage of the whole life cycle, and the main tasks of each stage are obviously different. Therefore, the life cycle of the hybrid energy storage system is firstly divided into three stages, and then cost analysis is carried out according to different characteristics of each stage. The life cycle staging of the hybrid energy storage system is shown in fig. 12.

The cost models for each stage are described in detail below.

1. Designing and constructing stage:

the design and construction stage mainly comprises the activities of project establishment, design, construction and the like, and the cost of the stage is generally called as one-time investment cost C_d. In order to simplify the calculation, the one-time investment cost is regarded as being generated at the starting point of the project, the one-time investment cost belongs to the current value of the capital and time value, and only the cost generated by the energy storage equipment is calculated.

C_d＝n_b,p·P_bat+n_b,c·C_bat+n_c,p·P_cap+n_c,c·C_cap

Wherein P is_bat，C_bat，P_cap，C_capRated power and rated capacity of the battery and the super capacitor respectively; n is_b,p,n_b,c,n_c,p,n_b,cThe unit price of power and the unit price of capacity of the battery and the super capacitor respectively.

2. Operation maintenance phase

The operation and maintenance stage mainly comprises normal operation and maintenance of the hybrid energy storage system and replacement of the battery pack and the super capacitor pack due to the end of life, and the costs generated by the two conditions are respectively called as maintenance cost C_mAnd replacement cost C_c. Maintenance cost C_mThe annual equal production belongs to the capital and time valueThe annual dollar of middle age; replacement cost C_cThe production of indefinite equal amount belongs to the fund outflow.

C_m＝N·(m_bat·C_bat+m_cap·C_cap)

Wherein m is_bat、m_capThe maintenance unit price of the battery and the super capacitor is shown; n is the number of the using periods of the equipment; q. q of_bat、q_capThe replacement times of the battery and the super capacitor are shown.

3. Scrap handling stage

The abandonment disposal stage mainly comprises the activities of removal, cleaning, destruction and the like after the service life of the hybrid energy storage system is expired, and the cost generated in the stage is called disposal cost C_s. The disposal cost two occurs at the end of the project and is the final value of the capital time value.

C_s＝l_b,p·P_bat+l_b,c·C_bat+l_c,p·P_cap+l_c,c·C_cap

Wherein l_b,p、l_b,c、l_c,p、l_c,cThe power disposal unit price and the capacity disposal unit price of the battery and the super capacitor.

Converting the costs into the form of present values to obtain:

the current value of the one-time investment cost is as follows:

C_d,p＝C_d

the current values of the maintenance cost are:

the current value of the replacement cost is:

the current value of disposal cost is:

here, the objective function of the capacity configuration of the hybrid energy storage system is set to be the annual average cost of the full life cycle cost, i.e., the cost

C＝min(C_d,p+C_m,p+C_c,p+C_s,p)/N

In configuring the hybrid energy storage system capacity, the following constraints should be considered:

(1) node power balance constraint:

wherein, P_iAnd Q_iIs the active and reactive power, U, injected by the node_iAnd U_jIs the voltage of the sum of nodes, delta_iAnd delta_jIs the phase angle of nodes i and j, G_ijAnd B_ijRespectively the real part and the imaginary part of the admittance matrix of the node, and N is the number of the nodes.

(2) And node power fluctuation constraint:

wherein, P_i ^maxAnd P_i ^minInjecting upper and lower limits of active power fluctuation, Q, for nodes_i ^maxAnd Q_i ^minAnd the upper limit and the lower limit of the reactive power fluctuation of the node are defined.

(3) Node voltage constraint:

U_i ^min≤U_i≤U_i ^max

wherein, U_i ^maxAnd U_i ^minThe upper and lower limits of the node voltage fluctuation.

(4) And (3) constraint of branch power flow:

wherein, P_l ^maxAnd P_l ^minInjecting upper and lower limits of active power fluctuation, Q, for nodes_l ^maxAnd Q_l ^minAnd the upper limit and the lower limit of the reactive power fluctuation of the node are defined.

(5) Constraint of conservation of energy. The sum of the wind power output, the battery output and the super capacitor output is consistent with the grid-connected power.

P_WG(t)+P_bat(t)+P_cap(t)＝P_out(t)

(6) And (5) constraint of confidence degree. The higher the confidence coefficient is, the better the grid-connected power performance is, the higher the cost is, and the cost and the confidence coefficient are often in a nonlinear relation, so that the fluctuation rate is not suitable economically. Therefore, a reasonable confidence level is selected to balance the technical index and the economic efficiency.

η≥η_min

(7) And (5) SOC constraint. The SOC of the battery and the SOC of the super capacitor are within a reasonable limit range, the service life of energy storage is shortened by overcharging and overdischarging, the SOC variation ranges of different types of energy storage media are different, and the SOC variation ranges are analyzed according to the adopted energy storage media.

Wherein: eta confidence degree of the output power of the energy storage system; k is a radical of_bat,min k_bat,max k_cap,min k_cap,maxThe minimum and maximum limit values of the SOC of the battery and the super capacitor are obtained.

For characteristic analysis of various energy storage batteries, the advantage of the all-vanadium redox flow battery is obvious, so that the all-vanadium redox flow battery is adopted as an energy type energy storage device in the embodiment, and the capacitor is adopted as a power type energy storage device. The capacity configuration method of the hybrid energy storage system in the previous section is applied to capacity configuration of renewable energy sources, and calculation is carried out based on a large amount of historical data of the wind power plant and the following condition parameters.

The various costs of the all-vanadium redox flow battery and the supercapacitor are shown in table 3.

TABLE 3 cost price per unit for two types of energy storage equipment

Type (B)	All-vanadium redox flow battery	Super capacitor
			Power unit price (Yuan/kW)	1500	1500
Capacity unit price (yuan/kWh)	1500	27000
			Maintenance unit price (Yuan/kWh)	5×10^-2	5×10^-2
Power disposal unit price (Yuan/kW)	120	60
			Capacity disposal Unit price (Yuan/kWh)	80	1000
Charge-discharge efficiency/%	80	95
			Upper limit of state of charge/%)	80	90
Lower limit of state of charge/%)	20	10

The predicted change curve of the annual average cost of HESS with the rated capacity of the battery is shown in fig. 13:

TABLE 4 optimization of the calculated results

The results of the optimization calculations are shown in table 4. According to the respective characteristics of energy storage and power storage, when the respective rated power is configured, the rated power value of the power storage, namely the super capacitor, is higher, and correspondingly, the energy storage capacity of the configuration of the energy storage, namely the battery energy storage, is larger. The calculation result conforms to the rule that the rated power of the battery is 16.33MW and the rated power of the stage capacitor is 8.73 MW. The maximum energy per day is mainly within 30MWh, so that in order to meet most energy requirements, the energy storage capacity of the battery is 28.76MWh, and the energy storage capacity of the super capacitor is 3.35 MWh.

In addition, in the case of the control group, only the all-vanadium redox flow battery was installed, and the same calculation was performed as shown in fig. 14.

The results of the optimization calculations are shown in table 5 below:

TABLE 5 optimization of the calculated results

Type (B)	Rated power (MW)	Rated capacity (MWh)	Annual average minimum cost (Wanyuan)
				All-vanadium redox flow battery	26.33	35.63	3514.4

It can be seen from the graph of fig. 14 that the annual average cost variation trends of the hybrid energy storage system and the battery-only energy storage system are consistent. The annual average cost is large when the battery capacity is small, and rapidly decreases with the increase of the rated capacity of the battery, and slowly increases when the battery capacity continues to increase. When the hybrid energy storage system is configured with a 16.33MW/28.76MWh battery, the super capacitor of 8.73MW/3.35MWh is, the annual average cost is minimum 3106.4 ten thousand yuan. When only the battery energy storage system is configured with 26.33MW/35.63MWh of battery, the minimum annual cost is 3513.4 ten thousand yuan. Therefore, the hybrid energy storage system reduces the capacity of a configured battery while increasing the configuration of the super capacitor, prolongs the service life of the energy storage system and reduces the cost.

According to the provisions of the technology for accessing the wind power plant to the power system, the maximum power fluctuation rate is divided into 1min interval power fluctuation rate and 10min interval power fluctuation rate, which both meet the limitation requirements of a power grid dispatching department, and the related fluctuation limitations are shown in table 6.

TABLE 6 wind farm maximum power fluctuation Standard

Installed capacity/MW of wind farm	Power fluctuation Limit/MW within 1min	Power fluctuation limit/MW within 10min
			<30	3	10
30-150	Installed capacity/10	Installed capacity/3
			>150	15	50

According to the principle that the wind power proportion exceeds the comprehensive target domain range, the wind power proportion is calculated by adopting the frequency repetition method probability of the overflow target domain, the proportion is compensated by a storage battery according to the energy storage response characteristic, and the proportion is compensated by a super capacitor, so that the respective running time, rated power and capacity ratio of the hybrid energy storage are determined.

And finally, the reasonability of the energy storage capacity configuration determined according to the minimum life cycle cost is verified through analysis of the wind power fluctuation condition. Therefore, the capacity of the energy storage system of the wind power plant is determined to be a battery of 16.33MW/28.76MWh and a super capacitor of 8.73MW/3.35MWh, and the method not only meets the effect of stabilizing wind power fluctuation, but also greatly reduces the investment cost.

And 3.1, calculating the relation between the energy storage capacity demand and the renewable energy source regulation.

And analyzing the change of the capacity of the energy storage system for stabilizing fluctuation of the photovoltaic power station along with the span of the data sample by using the current mature first-order digital low-pass filtering algorithm. Fig. 15 is a schematic diagram of a first-order digital low-pass filtering control algorithm. The controller collects the output power P of the photovoltaic module_sSetting the grid-connected power value P by a first-order digital low-pass filtering algorithm_out. Calculating the force output value of the energy storage system:

P_b＝P_out-P_s

required energy storage system capacity:

and selecting a photovoltaic power station output curve sample with the interval delta being 1min, continuously selecting the sample capacity from 1 day, accumulating the samples one by one, and calculating the capacity configuration result of the photovoltaic power station under the length of the sample.

Without loss of generality, the present invention randomly re-orders the contribution days 100 times and records the sample days required to reach the maximum capacity demand value after each re-ordering. On this basis, a histogram of frequency distribution of the number of days of the sample was plotted, and a normal distribution fitting was performed on the histogram, and the result is shown in fig. 16.

From the fitting frequency distribution effect in fig. 16, when the data span of the photovoltaic output is 29 to 32 days, the frequency of the maximum capacity demand is the maximum, that is, the probability of obtaining the required energy storage system capacity is the maximum when the sample days of 29 to 32 days are used. Conclusion in connection with section 3 the data span is 31 days when the allowable error accuracy is 0.25. Note that fig. 7 is distribution statistics of maximum capacity demand values and sample days under 100 random orderings, and therefore, the data span of the photovoltaic output data may be determined to be 31 days, and data of 4, 5, 6, 7, and 9 days may be randomly extracted from 5 weather categories of clear day, clear-to-cloudy day, cloudy-to-clear day, cloudy-to-cloudy day, cloudy-to-rainy day, and cloudy-to-cloudy day in sequence.

the output of the photovoltaic power station is related to natural conditions such as illumination intensity, environmental temperature and the like, and the control of the distribution rule of the power generation level of the photovoltaic power station on the time domain is beneficial to a dispatching department to reasonably arrange a power generation plan and reasonably select the span of data. Taking a certain photovoltaic power station as an example, the distribution condition of the annual output data of the photovoltaic power station in the time domain is quantitatively analyzed. The photovoltaic plant output level is defined as a percentage of installed capacity as shown in table 7.

TABLE 7 definition of photovoltaic power plant output levels

Level of output	Definition of
		High output	The output level is higher than 60 percent of the installed capacity
Medium output force	The output level is between 30% and 60% of the installed capacity
		Low output power	The output level is lower than 30 percent of the installed capacity

The photovoltaic power station output data characteristics comprise autocorrelation of the output data and similar daily clustering of the output data of the photovoltaic power station.

Autocorrelation of the force data.

The photovoltaic output data is considered as a time series. The relevance between the data value of the photovoltaic output data at any moment and the historical data can be inspected through the self-correlation characteristic analysis of the sequence, so that the data characteristics such as the implicit rule, the periodicity and the like in the sequence are found. The basis of the analysis of the autocorrelation properties is the correlation coefficient in statistics:

wherein cov (-) is covariance, var (-) is variance, p_xyIs a correlation coefficient, x_tData value at any moment, y, of the photovoltaic output data_tIs historical data. The correlation coefficient describes the consistency of the numerical variation trend among different sequences, and the larger the absolute value of the correlation coefficient is, the stronger the correlation of the variation among the sequences is. If one time series is shifted in time to obtain another time series, the correlation coefficient between the two time series is also called autocorrelation coefficient. The autocorrelation coefficients describe the consistency of the current data value with the data change before the time at which Δ t is the time shift.

The self-correlation coefficient shows a decaying oscillation trend along with the change of the time span, and the oscillation period of the self-correlation coefficient is 24 hours, which shows that the photovoltaic output has a very obvious regular change trend by taking 24 hours as the period. It is reasonable to determine the data span in the basic unit of 1 day data. With the continuous increase of the span, the oscillation amplitude of the autocorrelation coefficient is continuously reduced, and the oscillation center of the autocorrelation coefficient is gradually close to the point 0. After the span of more than 2400 hours (100 days), the maximum value of the autocorrelation coefficient of the photovoltaic sequence is reduced to be below 0.3, and the photovoltaic output at any moment is considered to have weak correlation with the change of photovoltaic historical data before 100 days, and the influence of the photovoltaic output is small. It was therefore initially concluded that it is preferable to take 100 days as a span for the analysis of the photovoltaic output data.

And (5) clustering similar days of the photovoltaic power station output data.

The analysis data further notes that a high output level corresponds to sunny days, a low output level corresponds to overcast and rainy days, and a medium output level includes three weather conditions of sunny to cloudy, cloudy to sunny, and cloudy. The analysis of the output level only describes the power generation capacity condition of the photovoltaic power station, but cannot describe the fluctuation distribution condition of the output of the photovoltaic power station. The generated energy and the fluctuation condition of photovoltaic output are strongly related to the weather condition. Typical forces in different weather are shown in fig. 17.

The photovoltaic output curve exhibits different shapes as the weather changes. The photovoltaic output of the same weather type contains similar information. And clustering the photovoltaic data with the same weather type. Selecting the solar irradiation intensity, the irradiation time and the air temperature as clustering indexes of photovoltaic output, namely daily characteristic vectors:

wherein, the first and the second end of the pipe are connected with each other,

the maximum temperature, the average temperature in the morning and the average temperature in the afternoon on the ith day; and T is the irradiation time length of the ith day.

Selecting photovoltaic data with span of 100 days for clustering analysis, and dividing the data into 5 types: sunny days; turning to cloudy days; turning to sunny cloud; rain and shade; cloudy. The clustering effect is shown in fig. 18:

as can be seen from fig. 18, there is a strong correlation between photovoltaic output and weather conditions. The photovoltaic output curve shapes under the same weather type are similar, namely the fluctuation situation is similar to the output level. The power output curves of the categories 1-3 also slightly fluctuate in a clear day period because a cloud layer shielding the sun occasionally appears in a clear day. Category 4 overcast rains differ from category 5 cloudiness in that: the solar irradiation is very small all day in rainy days, so the output amplitude is very small; the cloud is influenced by the movement of cloud layers, and the sun does not exist at present, so the amplitude of the output curve is larger than that of the output curve in rainy days, and the fluctuation is severe.

3.3, calculating the span based on the optimal sample capacity estimation;

the characteristics of the daily output data of the photovoltaic power station are reflected in the fluctuation intensity and the generated energy. Defining a solar output characterization coefficient B for characterizing the characteristics of photovoltaic output data on any day_i：

Wherein, P_ijIs the power value at the jth sample point on day i, and N is the total sample point per day. In order to determine the data span, the solar output characterization coefficients of 5 types of weather data, namely, sunny days, cloudy days and cloudy days, are respectively subjected to optimal sample volume estimation, and the sum of the sample volume estimations of all types is the data span. For a huge system state space, if the consequence analysis is performed on all system states to obtain an accurate result, the calculation process is often caused to be involved in a calculation disaster. Random sampling of the system state space should be used to replace the overall level with the samples taken. The more samples are drawn, the more comprehensive the reaction information. In practical application, however, the number of sampling points N cannot be too large, otherwise, the calculation amount is too large, and therefore, the error epsilon is allowed according to practical conditions₀And reasonably selecting the numerical value of N.

Is provided with (B)₁,B₂,B₃，B_n) Is a sample from population B, e (B) u, d (B) σ₂. From the central limit theorem for real numbers t_αThe method comprises the following steps:

it can be seen that, when the number of sample points n is sufficiently large,

the approximation follows a normal distribution N (0, 1). Thus, for a given confidence level α, there are:

wherein, t_αIs a bilateral quantile of standard normal distribution and can be obtained by looking up a normal distribution table. Let ε₀For the upper limit of the absolute error to be allowed, and for the sampling error to be reasonable, there should be:

comparing the two formulas, there are:

ε is the relative accuracy₀And = epsilon · u. The optimal sample volume is therefore:

where n is the sample volume, σ²Is the standard deviation of the sample, t_αIs the bilateral quantile of standard normal distribution, u is the overall mean, epsilon is the relative precision, and when the confidence coefficient is 95%, t_0.051.96, the overall mean u may be determined from the sample mean

And (6) estimating. In statistical theory, note that S²(S is a sample standard deviation) is σ²S may be used instead of sigma. The relative accuracy is set by the investigator, and is herein taken to be 0.15, 0.2, 0.25, 0.3, respectively. The most suitable sample capacity estimation at different allowable error accuracies is shown in fig. 19:

as can be seen from fig. 19, as the allowable error precision becomes smaller, the optimal sample capacity estimation for each weather category increases, and the overall data span increases continuously. After photovoltaic output data are clustered, when the allowable error precision is fixed, X is randomly extracted from 5 weather categories (sunny day 1; sunny to cloudy day 2; cloudy to sunny 3; cloudy to cloudy day 4; cloudy to cloudy day 5) in sequence_iData of (i ═ 1, 2, 3, 4, 5) days, that is, dataSpan.

The output level and the fluctuation characteristic of photovoltaic power generation data are known.

It should be emphasized that the embodiments described herein are illustrative rather than restrictive, and thus the present invention is not limited to the embodiments described in the detailed description, but also includes other embodiments that can be derived from the technical solutions of the present invention by those skilled in the art.

Claims

1. The multi-element composite energy storage optimal configuration method for supporting large-scale renewable energy consumption is characterized by comprising the following steps of: the method comprises the following steps:

2. The multi-element composite energy storage optimal configuration method supporting large-scale renewable energy consumption according to claim 1, characterized in that: the step 1 comprises the following steps:

3. The multi-element composite energy storage optimal configuration method supporting large-scale renewable energy consumption according to claim 1, characterized in that: the step 1.1 comprises the following steps:

step 1.1.1, distributing energy of a hybrid system based on low-pass filtering;

the constraint conditions are as follows:

b₁≤λ(t)≤b₂

wherein, P_bat,eIs the rated power of the battery, C_soc,capIs SOC, C of a super capacitor_soc,medAt moderate level of SOC, b₁、b₂For factory-set parameters, M_capIs the energy storage capacity, Delta, of a supercapacitort is the charging and discharging time interval, and the constructed model is calculated by NSGA/algorithm, and a control strategy is obtained: dividing a command period of the output of the hybrid energy storage system into two parts, wherein in the ith command period T_iThe latter part of the period T_iWithin 2, at T_i2, ensuring the SOC of the super capacitor to be at a certain level, and selecting f₂The better solution is taken as the current solution; when the (i + 1) th instruction cycle is entered, the previous part of the cycle is T_i+1Within +1, the solution selected to obtain the better solution is taken as the current solution.

4. The multi-element composite energy storage optimal configuration method supporting large-scale renewable energy consumption according to claim 2, characterized in that: the energy storage planning model of the step 1.2 is as follows:

the charging process is as follows:

C_SOC,bat(t)＝C_SOC,bat(t-Δt)+P_bat(t)Δtη_bat,c/C_bat

C_SOC,cap(t)＝C_SOC,cap(t-Δt)+P_cap(t)Δtη_cap,c/C_cap

the discharging process is as follows:

C_SOC,bat(t)＝C_SOC,bat(t-Δt)+P_bat(t)Δt/(η_bat,d·C_bat)

C_SOC,cap(t)＝C_SOC,cap(t-Δt)+P_cap(t)Δt/(η_cap,d·C_cap)

the constraint conditions are as follows:

|P_bat(t)|≤|P_bat,lim(t)|

|P_cap(t)|≤|P_cap,lim(t)|

wherein, C_SOC,bat(t) and C_SOC,cap(t) is the change condition of the SOC of the two energy storage systems; Δ t is the sampling interval; eta_bat,cThe charging efficiency of the battery; eta_cap,cIs the discharge efficiency of the cell; eta_bat,dThe charging efficiency of the super capacitor is improved; eta_cap,dIs the discharge efficiency of the supercapacitor; c_batIs the capacity of the battery; c_capThe capacity of the super capacitor; i P_bat,lim(t) | is the maximum charging power value allowed by the battery at the time t; i P_cap,lim(t) | is the maximum charging power value allowed by the super capacitor at the time t;

and (3) discharging:

wherein, P_bat,cmaxThe maximum charging power value determined by the self characteristics of the battery; p is_cap,cmaxThe maximum discharge power value is determined by the self characteristics of the battery; p_bat,dmaxThe maximum charging power value is determined by the self characteristics of the super capacitor; p_cap,dmaxThe maximum discharge power value is determined by the self characteristics of the super capacitor; c_SOC,bat,maxAn upper limit of the SOC constraint of the battery; c_SOC,cap,maxA lower constraint limit for the SOC of the battery; c_SOC,bat,minAn SOC constraint upper limit for the supercapacitor; c_SOC,cap,minThe SOC constraint lower limit of the super capacitor.

5. The multi-element composite energy storage optimal configuration method supporting large-scale renewable energy consumption according to claim 1, characterized in that: the step 2 comprises the following steps:

step 2.1, calculating the equivalent cycle life of the battery;

calculating the depth of discharge of the battery by a rain flow counting method, and calculating the equivalent cycle life of the battery according to the corresponding relation between the depth of discharge and the cycle life of the battery;

step 2.2, calculating the rated power range of the energy storage system;

step 2.3, calculating the capacity range of the energy storage system;

step 2.4, calculating cost;

6. The method for optimally configuring the multi-element composite energy storage supporting large-scale renewable energy consumption according to claim 5, wherein the method comprises the following steps: the cost calculated in the step 2.4 comprises the life cycle cost, the time value of capital, the construction stage of engineering and constraint conditions;

the method for calculating the life cycle cost comprises the following steps: the life cycle cost of the hybrid energy storage system is mainly concentrated on the investment cost of the battery and the super capacitor module, the operation and maintenance cost after the system is operated, and the disposal cost after scrapping; the economic evaluation of the energy storage cost is influenced by the selection factors of the load characteristics, the parameters of the energy storage system and the working mode, and the economic evaluation of the energy storage system is carried out according to the typical situation of the applicable power system and the technical specifications and parameter requirements of the energy storage system defined according to various power applications during calculation;

the time value of the fund is calculated by the following method:

C_d＝n_b,p·P_bat+n_b,c·C_bat+n_c,p·P_cap+n_c,c·C_cap

C_m＝N·(m_bat·C_bat+m_cap·C_cap)

wherein m is_batAnd m_capMaintenance unit prices of the battery and the super capacitor respectively; n is the number of the using periods of the equipment; q. q.s_batAnd q is_capThe replacement times of the battery and the super capacitor are respectively;

C_s＝l_b,p·P_bat+l_b,c·C_bat+l_c,p·P_cap+l_c,c·C_cap

wherein l_b,p、l_b,c、l_c,pRespectively areThe power handling unit price and the capacity handling unit price of the battery and the super capacitor;

7. The multi-element composite energy storage optimal configuration method supporting large-scale renewable energy consumption according to claim 1, characterized in that: the energy storage optimization configuration in the step 2.5 is as follows:

according to the principle of wind power ratio exceeding the comprehensive target domain range, the wind power ratio is counted by adopting the frequency repetition method of the overflow target domain, the wind power ratio is compensated by a storage battery according to the energy storage response characteristic, the wind power ratio with large ratio is compensated by a super capacitor, and the respective running time, rated power and capacity ratio of the hybrid energy storage are determined.

8. The multi-element composite energy storage optimal configuration method supporting large-scale renewable energy consumption according to claim 1, characterized in that: the step 3 comprises the following steps:

the controller collects the output power P of the photovoltaic module_sSetting the grid-connected power value P by a first-order digital low-pass filtering algorithm_outAnd calculating the force output value of the energy storage system:

P_b＝P_out-P_s

the required energy storage system capacity is:

wherein the content of the first and second substances,

the maximum value, the average value in the morning and the average value in the afternoon of the solar irradiation intensity on the ith day; t is t_hi

The maximum temperature, the average temperature in the morning and the average temperature in the afternoon on the ith day; t is the irradiation time length of the ith day; according to the autocorrelation of the output data, selecting photovoltaic data with the span of 100 days for clustering analysis to obtain strong correlation between the photovoltaic output and weather conditions;

3.3, calculating the span based on the optimal sample capacity estimation;

the optimal sample volume is: